A document scanner receives light reflected from a document, and converts the image on the document into a format for further processing. An analog light lens scanner uses a lens to project the image onto an apparatus, such as a photoreceptor, to reproduce the document image through electrostatic printing. Scanners may also use digital photosensors that convert the light reflected from the document into digital electronic information. A lens may focus the document image onto the digital photosensors. Alternatively, a linear array of photosensors may extend across the entire width of the document to be scanned. The linear array of photosensors may be moved along the document, or the document may be moved past the linear array of photosensors. In certain instances, it may be possible to scan the documents using digital photosensors that move in a raster format, moving the photosensors across the document in a predetermined pattern.
For many document image processing functions, it is useful to identify the precise location of the edge of the document. For example, in a scanning function, identifying the edges of the document permits the scanner to determine the size of the document being scanned, and its position. Knowing the size and position of the document also permits the scanner to identify the document background, so that any image processing functions performed by the scanner can compensate for or otherwise appropriately handle particular document backgrounds, such as color backgrounds.
Various edge detection algorithms and techniques are currently in use and well understood in the art. Typically, such techniques identify edge-like features that indicate possible document edges, and use the difference in background intensity between the document and the scanner's document backer to distinguish the document from the backer. Such differences may be difficult to detect if the document background is similar to the scanner's document backer. In addition, a frequent challenge for an edge detection algorithm is to avoid identifying marks or other variations on the document backer as a document edge. Another challenge is to avoid identifying data from a poorly functioning scanner detector as a document edge.